+++
date = "2014-09-05T00:00:00+01:00"
title = "On open source"
+++
Last week I read a great article
http://videlalvaro.github.io/2014/08/on-contributing-to-opensource.html[on
contributing to open source] by Alvaro Videla. He makes
many great points and I am in agreement with most of it.
This made me want to properly explain my point of view with
regard to open source and contributions. Unlike most open
source evangelism articles I will not talk about ideals or
any of that crap, but rather my personal feelings and
experience.
I have been doing open source work for quite some time.
My very first open source project was a graphics driver
for (the very early version of) the PCSX2 emulator. That
was more than ten years ago, and there
http://ngemu.com/threads/gstaris-0-6.30469/[isn't
much left to look at today]. This was followed by a
https://github.com/extend/wee[PHP framework]
(started long before Zend Framework was even a thing) and
a few other small projects. None of them really took off.
It's alright, that's pretty much the fate of most open
source projects. You spend a lot of work and sweat and
get very little in return from others.
This sounds harsh but this is the reality of all open
source projects. If you are thinking of building a project
and releasing it as open source, you should be prepared
for that. This is how most of your projects will feel like.
Don't release a project as open source thinking everyone
will pat you on the back and cheer, this won't happen. In
fact if your project is a too small improvement over existing
software, what many people will do is say you have NIH
syndrome, regardless of the improvement you bring. So you
need not to rely on other people in order to get your
enjoyment out of building open source software.
In my case I get enjoyment from thinking about problems
that need solving. Often times the problems are already
solved, but nevermind that, I still think about them and
sometimes come up with something I feel is better and then
write code for it. Writing code is also fun, but not as
fun as using my brain to imagine solutions.
You don't need thousands of users to do that. So are
users worthless to me then? No, of course not. In fact
they are an important component: they bring me problems
that need solving. So users are very important to me.
But that's not the only reason.
I got lucky that the Cowboy project became popular.
And seeing it be this popular, and some of my other projects
also do quite well, made me believe I could perhaps work
full time on open source. If I can work full time then
I can produce better software. What I had one hour to
work on before I can now spend a day on, and experiment
until I am satisfied. This is very useful because that
means I can get it almost right from the beginning, and
avoid the million API breaking changes that occured
before Cowboy 1.0 was released.
To be able to work full time on open source however,
I need money. This is a largely unspoken topic of open
source work. The work is never free. You can download the
product for free, but someone has to pay for the work
itself. Life is unfortunately not free.
Large projects and some lucky people have their work
sponsored by their employers. Everyone else has to deal
with it differently. In my case I was sponsored for a
while by the http://leo-project.net/leofs/[LeoFS]
project, but that ended. I also had the Farwest fundraiser,
which was a success, although the project stalled after that.
(Fear not, as Farwest will make a comeback as a conglomerate
of Web development projects in the future.) After that I set
up the http://ninenines.eu/support/[sponsoring scheme],
which I can proudly say today brings in enough money to
cover my food and shelter. Great!
This is a start, but it's of course not enough. Life
is a little more than food and shelter, and so I am still
looking for sponsors. This is not a very glorious experience,
as I am essentially looking for scraps that companies can
throw away. Still, if a handful more companies were doing
that, not only would I be able to live comfortably, but I
would also be able to stop worrying about the future as I
could put money on the side for when it gets rough.
A few companies giving me some scrap money so I could
live and work independently is by far the most important
thing anyone can do to help my projects, including Cowboy.
Yes, they're even more important than code contributions,
bug reports and feedback. Because this money gives me the
time I need to handle the code contributions, bug reports
and feedback.
If Cowboy or another project is a large part of your
product or infrastructure, then the best thing you can do
is become a sponsor. The second best is opening tickets
and/or providing feedback. The third best is providing
good code contributions.
I will not expand on the feedback part. Feedback is
very important, and even just a high five or a retweet
is already good feedback. It's not very complicated.
I want to expand a little on code contributions
however. Not long ago I ran across the term "patch bomb"
which means dropping patches and expecting the project
maintainers to merge them and maintain them. I receive
a lot of patches, and often have to refuse them. Causes
for refusal vary. Some patches only benefit the people
who submitted them (or a very small number of people).
Some patches are not refined enough to be included.
Others are out of scope of the project. These are some
of the reasons why I refuse patches. Having limited
time and resources, I have to focus my efforts on the
code used by the larger number of users. I have to
prioritize patches from submitters who are reactive
and address the issues pointed out. And I have to plainly
refuse other patches.
I believe this wraps up my thoughts on open source.
Overall I had a great experience, the Erlang community
being nice and understanding of the issues at hand in
general. And if the money problem could be solved soon,
then I would be one of the luckiest and happiest open
source developer on Earth.
Think about it the next time you see a donation button
or a request for funds or sponsoring. You can considerably
improve an open source developer's life with very little
of your company's money.